📝 Study asset for the dbt Analytics Engineering Certification Practice Tests course on Udemy — 195 exam-style questions across 3 timed tests, updated for dbt 1.11.
| Detail | Value |
|---|---|
| Duration | 2 hours (120 minutes) |
| Questions | 65 |
| Passing score | 65% (need 42+ correct out of 65) |
| Price | $200 USD |
| Format | Online, live proctored via Zoom |
| Languages | English, Japanese |
| dbt version | 1.11 (renewed April 2026) |
| Results | Immediate — score shown right after you submit |
| Certification validity | 2 years from award date |
| Retakes | Allowed (new $200 fee required) |
| Cancellation | Free up to 24 hours before. No refund for no-shows. |
Register at: pages.talview.com/dbtlabs/certifications
After registering, you schedule a time slot. The exam is administered via Talview with a live Zoom proctor.
This is where the dbt exam differs from typical Udemy-style tests. The real exam uses 6 different question formats:
Standard format — 4 options, pick 1 correct answer.
Example: Which materialization stores the transformation result as a physical table in the data warehouse? A) view B) table C) ephemeral D) incremental Answer: B
Pick 2 or 3 correct answers from 4-5 options. The question always tells you how many to select.
Example: Which TWO of the following are valid snapshot strategies in dbt? (Choose 2) A) timestamp B) merge C) check D) append Answer: A, C
You type the exact answer. No options to choose from. Syntax must be precise.
Example: What dbt command installs the packages listed in your
packages.ymlfile? Answer:dbt deps
Match items from two columns. For example, match dbt commands to their descriptions, or match error types to their causes.
You're shown a code snippet or DAG diagram and must click on the exact location that answers the question. For example:
- "Click on the line that causes this compilation error"
- "Click on the model in this DAG that would run first"
This tests precision — you can't just recognize the concept, you must identify the exact location.
The hardest format. Options are shown one at a time. For each option, you must decide YES or NO — without seeing the other options. This means:
- You can't use process of elimination
- You can't compare options
- You must know the answer confidently for each option individually
- It's psychologically designed to test genuine knowledge vs. educated guessing
Example (you see one option at a time): Question: "Which of the following are valid dbt materializations?" Option 1: "table" → Yes/No? Option 2: "view" → Yes/No? Option 3: "cached" → Yes/No? Option 4: "incremental" → Yes/No?
- 1 point per correct answer
- 0 points for incorrect answers (no negative marking)
- All questions weighted equally
- Unscored research questions are mixed in — some questions don't count toward your score, but you can't tell which ones they are. Answer every question seriously.
Every question maps to one of these 7 v1.11 domains (the renewed April 2026 blueprint collapsed Documentation into Topic 1):
| # | Domain | What It Covers |
|---|---|---|
| 1 | Developing and optimizing dbt models | Materializations (including microbatch with event_time/begin/batch_size/lookback), ref/source, incremental strategies (append/merge/delete+insert/insert_overwrite), snapshots (YAML-based), seeds, Python models, Jinja/macros, grants, DAG structure, dbt_project.yml, persist_docs, doc blocks, --empty and --sample dry-run flags |
| 2 | Managing dbt models governance | Contracts (enforced: true, required data_type per column), YAML column-level constraints (primary_key/not_null/check), versions and deprecation_date |
| 3 | Debugging data modeling errors | Error types (Compilation/Database/Runtime), compiled SQL debugging, YAML errors, Jinja errors, behavior-change flags in dbt_project.yml (e.g. require_explicit_package_overrides_for_builtin_materializations) |
| 4 | Troubleshooting and optimizing dbt pipelines | dbt clone, DAG failure management, integrated tool troubleshooting, dbt retry |
| 5 | Implementing dbt tests | Generic/singular/custom/custom generic tests, test severity, store_failures, unit tests, testing workflows |
| 6 | Implementing and Maintaining External Dependencies | Exposures (type, maturity, depends_on, owner), source freshness (loaded_at_field, warn_after, error_after), packages.yml |
| 7 | Leveraging the dbt state | State artifacts (manifest.json), state:modified/state:new selectors, result selectors, --defer, --state, Slim CI, dbt retry |
- You join a Zoom call with a live proctor
- The proctor verifies your identity (government-issued ID)
- You share your screen throughout the exam
- You cannot use any notes, books, or second screens
- You cannot leave the camera frame
- The exam interface shows questions one at a time with a progress bar
- You can flag questions and return to them later
- A timer counts down from 2:00:00
| Phase | Time | Action |
|---|---|---|
| First pass | 0:00 - 1:30 | Answer all questions you're confident about. Flag uncertain ones. (~1.5 min/question) |
| Second pass | 1:30 - 1:50 | Return to flagged questions. Use remaining time to think carefully. |
| Final check | 1:50 - 2:00 | Quick scan — make sure nothing is blank. Trust your first instinct for guesses. |
Pace: 65 questions in 120 minutes = 1 minute 50 seconds per question. Most straightforward questions take 30-60 seconds. Budget extra time for code-reading and DOMC questions.
The exam was significantly renewed in April 2026 to align with dbt Core 1.11. Key changes from the v1.7 (May 2024) blueprint:
| Added | Removed/Reduced |
|---|---|
Microbatch incremental materialization (event_time, begin, batch_size, lookback) |
access: public/protected/private (out of scope) |
--empty flag (zero-row dry-run) |
groups (out of scope) |
--sample flag (temporal-subset dry-run) |
Documentation as a standalone domain (folded into Topic 1) |
Behavior-change flags in dbt_project.yml |
Some Documentation depth (lineage / persist_docs are still in scope under Topic 1) |
YAML column-level constraints (primary_key/not_null/check) and the data_type requirement when contracts are enforced |
|
Explicit incremental-strategy selection by dataset (append vs merge vs delete+insert vs insert_overwrite) |
|
dbt clone (now under Topic 4) |
If you're using study materials from before April 2026, they're outdated. Make sure your prep covers the new topics — microbatch, --empty/--sample, behavior flags, YAML constraints, and incremental-strategy selection.
- Rushing through DOMC questions — These require the most careful thought. Slow down.
- Not reading code carefully — Hotspot questions require identifying the exact problematic line.
- Second-guessing yourself — The exam adds deliberate extra details to make you doubt your answer. Trust your knowledge.
- Skipping "easy" questions — Unscored research questions can look easy. Answer everything.
- Running out of time — Flag hard questions and come back. Don't spend 5 minutes on one question.
- Not memorizing YAML syntax — Fill-in-the-blank questions require exact syntax.
warn_after:, notwarning_after:.
- Immediate results — Your score is displayed right after submission
- Pass: You receive a digital certificate and badge within a few days
- Fail: You can retake immediately (new $200 fee). Your score report shows which domains were weak.
- Certification is valid for 2 years — After that, you must recertify (new exam, new fee)
- Share it: Add the certification to your LinkedIn profile. The digital badge links to verification.
Updated to the 2026 dbt v1.11 Certification